With the economic environment in Hong Kong, it is a very difficult year in 2019 for SDi. However, we are a consulting company with long term commitment to our staff and clients. Thus, we are investing a significant amount to enhance our computer lab with GPU and new servers with SSD.
Data Science Cloud (Private Cloud)
For the private cloud for Data Science, it is basically referenced on Archsolution – Private Cloud. Their cloud platform is a very good answer for those cloud vendors charging high on data storage, database or even Big Data platform. It is not a good idea for paying high costs on tons of complex calculations and 50Tb data processing being done in any cloud vendor for 7×24 operations.
With the help of Archsolution Limited, we are using Gigabyte server / workstation mainboards for custom server building. Our architecture diagram for the private data science cloud as shown as below.
(NOTE: However, the diagram is just showing the Hadoop but not the complete picture for our analytic and machine learning platform.)
Figure 1: Open Stack architecture – part of our data science cloud
To build the private data science cloud, it is aimed to have our infrastructure without vendor lock-in for research and training purposes. There are some vital R&D projects like IoT and AI running in this platform.
For HADR, it is just needed to duplicate 1 set to another data center with the synchronization to maintain the information stored at near real-time basis.
For the management of Open Stack cloud, we use chef and compass to maintain the daily operation for the private cloud. In a nutshell, it is not only related to cost saving but more vital as the full control for a learning company like us.
Development Workstation
For development workstation, it is important to categorize different types of workload like predictive analytic and Machine Learning (ML).
If there is something related to AI/ ML, it is recommended to have a workstation installed at least one powerful Graphic Processing Unit (GPU). In order to save the cost, we are taking refurbished gaming PC running Intel i7 or Xeon CPU with different GPU like the Nvidia GTX 2070.
However, it is better to have detailed comparison between the speed and performance between GPU with the URL below:
Deep Learning GPU Benchmarks – Tesla V100 vs RTX 2080 Ti vs GTX 1080 Ti vs Titan V: https://lambdalabs.com/blog/best-gpu-tensorflow-2080-ti-vs-v100-vs-titan-v-vs-1080-ti-benchmark/
GPU Server
The technology giant – Google is using their own custom-made hardware to run TensorFlow for AI and machine learning. However, there is only Huawei providing AI specialized hardware in the public market. In order to have an alternative highly available, it is suggested to have a hardware box running a couple of fastest GPU units like NVIDIA 2080Ti (in gaming market) or NVIDIA Tesla / Titan (officially recommended) to do the intensive calculations.
Unfortunately, our data science team is still using a box of containing 6 pieces of Nvidia GTX 1080Ti as a retired Ethereum miner. However, we are planning to upgrade the cluster with the memory size and speed concerns. It is suggested to have 32GB ram cards like Titan V100 for handling a huge Machine Learning / Deep Learning model under the TensorFlow framework.
Figure 2: our Ethereum miner with 6 GPU (1080 Ti)
We are also trying to use anther Ethereum miner with 6 pieces of RX570 to see whether it is effective to do machine learning work-load.
For the investment, it is definitely not a small amount but it is our long-term commitment to our staff and our client with the continuous development on the team’s skill set. Basically, we are proud to have team members able to face different challenges by the development of data science skill-set rather than using tools to copy other people applications.
